Glucocerebrosidase reduces the spread of protein aggregation in a Drosophila melanogaster model of neurodegeneration by regulating proteins trafficked by extracellular vesicles

Summary

Abnormal protein aggregation within neurons is a key pathologic feature of Parkinson’s disease (PD). The spread of brain protein aggregates is associated with clinical disease progression, but how this occurs remains unclear. Mutations in glucosidase, beta acid 1 (GBA), which encodes glucocerebrosidase (GCase), are the most penetrant common genetic risk factor for PD and dementia with Lewy bodies and associate with faster disease progression. To explore how GBA mutations influence pathogenesis, we previously created a Drosophila model of GBA deficiency (Gba1b) that manifests neurodegeneration and accelerated protein aggregation. Proteomic analysis of Gba1b mutants revealed dysregulation of proteins involved in extracellular vesicle (EV) biology, and we found altered protein composition of EVs from Gba1b mutants. Accordingly, we hypothesized that GBA may influence pathogenic protein aggregate spread via EVs. We found that accumulation of ubiquitinated proteins and Ref(2)P, Drosophila homologue of mammalian p62, were reduced in muscle and brain tissue of Gba1b flies by ectopic expression of wildtype GCase in muscle. Neuronal GCase expression also rescued protein aggregation both cell-autonomously in brain and non-cell-autonomously in muscle. Muscle-specific GBA expression reduced the elevated levels of EV-intrinsic proteins and Ref(2)P found in EVs from Gba1b flies. Perturbing EV biogenesis through neutral sphingomyelinase (nSMase), an enzyme important for EV release and ceramide metabolism, enhanced protein aggregation when knocked down in muscle, but did not modify Gba1b mutant protein aggregation when knocked down in neurons. Lipidomic analysis of nSMase knockdown on ceramide and glucosylceramide levels suggested that Gba1b mutant protein aggregation may depend on relative depletion of specific ceramide species often enriched in EVs. Finally, we identified ectopically expressed GCase within isolated EVs. Together, our findings suggest that GCase deficiency promotes accelerated protein aggregate spread between cells and tissues via dysregulated EVs, and EV-mediated trafficking of GCase may partially account for the reduction in aggregate spread.

Keywords:

Actins – Biosynthesis – Drosophila melanogaster – Muscle proteins – Muscle tissue – Parkinson disease – Principal component analysis – Protein aggregation

Introduction

Parkinson’s disease (PD) is the most common neurodegenerative movement disorder, affecting 1–2% of people over 65 years of age [1]. PD is characterized by cardinal motor and non-motor symptoms, including rigidity, slowness of voluntary movements, and cognitive decline [2–4]. Intraneuronal Lewy bodies containing ubiquitinated proteins and α-synuclein are a hallmark pathologic finding in PD. The stereotypic spread of Lewy bodies in PD from the rostral brain stem to the midbrain and eventually throughout the neocortex suggests a prion-like mechanism mediating propagation of protein aggregates from neuron to neuron [5]. This temporo-spatial spread of Lewy bodies correlates with clinical progression of PD and has been replicated in several animal models [6–8]. Although much work has focused on identifying genes involved in PD, and how perturbation of these genes lead to PD pathogenesis, the mechanisms underlying PD are not yet completely understood.

Mutations in the gene glucosidase, beta acid 1 (GBA), encoding the lysosomal ceramide metabolism enzyme glucocerebrosidase (GCase), are the strongest genetic risk factor for PD and dementia with Lewy bodies, increasing risk of developing PD by approximately 5-fold in GBA mutation carriers compared to non-carriers [9–11]. GBA carriers with PD are otherwise clinically similar to idiopathic PD patients, with indistinguishable response to dopaminergic medications, slightly younger age of onset by about 4 years, and higher incidence of cognitive decline [12,13]. While some studies suggest that GBA carriers with PD may have a heavier burden of Lewy bodies than in non-carriers with PD, the neuropathologic features are similar [14]. Importantly, mutations in the gene GBA are common, having been found in 4–5% of all idiopathic PD patients [13,15]. Recent longitudinal clinical studies have revealed that in addition to increased risk, GBA carriers with PD have faster progression of both motor and cognitive symptoms compared to idiopathic PD patients [16–18].

Several models have been developed and characterized using Drosophila melanogaster to examine how GBA influences PD pathogenesis [19–22]. Our previously characterized Drosophila model of GBA deficiency (Gba1b) manifests several phenotypes reminiscent of key features of PD, including neurodegeneration, locomotor deficits, cognitive deficits, and accelerated protein aggregation in multiple tissues including the nervous system and muscle [19]. Proteomic analysis of our Gba1b mutant flies revealed that proteins involved in extracellular vesicle (EV) biology were dysregulated, and EVs isolated from Gba1b Drosophila hemolymph revealed increased levels of aggregate-prone and EV-intrinsic proteins, indicating that GBA deficiency alters the protein composition of EVs [23]. EVs are a heterogeneous group of membrane-bound vesicles secreted by cells that can have multiple functions, including intercellular communication through protein, lipid, and nucleic acid cargo and discard of cell components outside of the cell [24,25]. EVs can originate from the endosomal system as a result of fusion of the late endosome to the plasma membrane, releasing intraluminal vesicles as EVs into the extracellular matrix (exosomes), or from direct outward budding of the plasma membrane (microvesicles) [24,26,27]. EVs have been implicated in the propagation of misfolded proteins between cells in multiple neurodegenerative diseases, including PD [28–33]. α-synuclein has been found within EVs isolated from tissues of PD and dementia with Lewy bodies patients [28,34], and in vitro studies have suggested that α-synuclein within EVs may be more pathogenic than free extracellular α-synuclein [35]. Accordingly, we hypothesized that tissue-specific expression of wildtype (WT) GCase might reduce the protein aggregates that accumulate in Gba1b mutants.

Using our GBA-deficient Drosophila model [18,23], we examined whether GCase could be mediating propagation of protein aggregates from cell-to-cell via EVs. In this study, we found that tissue-specific expression of WT GCase in Gba1b mutants corrected the alterations in protein composition of GBA-deficient EVs, as well as protein aggregation in local and distant tissues. Perturbing an EV biogenesis pathway in Gba1b mutants by knocking down neutral sphingomyelinase (nSMase), encoding an enzyme important for EV release as well as ceramide metabolism, resulted in tissue-specific cell-autonomous effects on protein aggregation and changes in EV protein composition but did not rescue protein aggregation in distant tissues. Lipidomic analysis of the effect of nSMase knockdown on ceramide (Cer) and glucosylceramide (GlcCer) levels suggested that Gba1b mutant protein aggregation may be dependent on relative depletion of specific Cer species that are known to be enriched in EVs. Finally, we observed ectopically expressed GCase in EVs, suggesting that trafficking of GCase within EVs may contribute to the observed non-cell-autonomous rescue. Our findings suggest that mutations in GBA result in alterations in the lipid composition of cells and the protein composition of EVs that promote enhanced cell-to-cell transmission of pathogenic protein aggregates. Moreover, our findings indicate that GCase can be packaged into EVs and trafficked between cells to reduce protein aggregation throughout an organism. These findings suggest a possible mechanism underlying the clinical finding that GBA mutation carriers not only have increased risk of developing PD, but also faster progression of disease.

Results

Protein aggregation in Gba1b mutants can be rescued cell-autonomously and non-cell-autonomously by tissue-specific dGba1b expression

Our prior work revealed accelerated aggregate accumulation and dysregulation of EV-related proteins in Gba1b mutant flies suggesting that GCase deficiency may influence cell-to-cell spread of protein aggregates. To explore the hypothesis that mutations in GBA promote the spread of protein aggregates from peripheral tissues to the brain, we expressed WT Drosophila Gba1b (dGba1b) in non-neural tissues of Gba1b mutants and examined accumulation of ubiquitinated protein and Ref(2)P, the Drosophila homologue of mammalian p62, in the brain. Using the DMef-GAL4 driver to drive expression of WT dGba1b in muscles throughout the fly, insoluble ubiquitinated protein aggregate accumulation was reduced in both the thoraces and heads of Gba1b mutant flies (Fig 1A and 1B). This non-cell-autonomous rescue of ubiquitinated protein aggregates was dramatically apparent in the whole brains of Gba1b mutants (Fig 1C). However, because DMef-GAL4 is also expressed in muscles located in the fly head that control the proboscis and may not exclusively drive expression in only muscle cells, we repeated this experiment using the Act88F-GAL4 driver. Act88F expression is specifically in the thoracic indirect flight muscles and is not found in the head [36]. Expression of WT dGba1b using the Act88F-GAL4 driver in Gba1b mutants significantly reduced accumulation of insoluble ubiquitinated proteins in the thoraces and heads (Fig 1E and 1F) and rescued their shortened lifespan (Fig 1D and S1 Table).

Ref(2)P, the Drosophila ortholog of mammalian SQSTM1/p62, is important for selectively targeting ubiquitinated proteins for lysosomal degradation. While Ref(2)P is often used as a marker of autophagic flux, we have previously found it to accumulate in Gba1b mutants with no evidence of significantly impaired autophagy [23]. However, Ref(2)P is also required for aggregate formation under normal physiological conditions [37], and it has been shown to be released concomitantly with α-synuclein in EVs [34]; therefore we used its accumulation as a second readout of protein aggregation. We anticipated that Ref(2)P accumulation in Gba1b mutants could also be non-cell-autonomously rescued. We indeed found that Ref(2)P accumulation was significantly decreased in both the thoraces and heads of Gba1b mutants expressing WT dGba1b in flight muscle (Fig 1G and 1H).

To determine whether there might be a directionality or tissue-specificity to the non-cell-autonomous rescue of protein aggregation, we expressed WT dGba1b in the nervous system and assessed for protein aggregates in the body. Neuronal expression of WT dGba1b using the elav-GAL4 driver decreased both Ref(2)P accumulation and insoluble ubiquitinated proteins in both the heads and bodies of Gba1b mutants (Fig 2A–2D). In addition, neuronal expression of WT dGba1b partially rescued the shortened lifespan of Gba1b mutant flies (Fig 2E and S2 Table). Our findings indicate that GBA has both cell-autonomous and non-cell-autonomous roles in preventing protein aggregation. The non-cell-autonomous role of GBA raises the possibility that GCase may normally function to reduce the spread of protein aggregates from one cell and tissue to another.

Non-cell-autonomous rescue of protein aggregation in Gba1b mutants is mediated by extracellular vesicles

There are multiple mechanisms that could mediate non-cell-autonomous interactions, including exocytosis of cytoplasmic components into the extracellular matrix, direct cell-to-cell contacts, and release of cytoplasmic components via EVs. Given that our prior proteomic analysis of Gba1b mutants found evidence of dysregulation of EVs [23], we hypothesized that GCase influences the trafficking of aggregate-prone proteins in EVs. We isolated EVs from hemolymph and confirmed that Ref(2)P was increased in EVs from Gba1b mutants compared to controls (Fig 3A), as we had seen previously [23]. We also found that EV-intrinsic proteins Rab11 and Rab7 were elevated, although the increase in Rab11 did not reach statistical significance (Fig 3B and 3C). We tested the possibility that muscle-specific expression of WT dGba1b would reduce the increased levels of Ref(2)P and EV-intrinsic proteins from Gba1b mutants. Indeed, we observed a reduction in Ref(2)P and Rab7 levels in EVs isolated from Gba1b mutants expressing WT dGba1b in flight muscle using the Act88F-Gal4 driver (Fig 3A and 3C). However, while ubiquitinated proteins were increased in Gba1b mutant whole flies, they were not significantly altered in EVs isolated from Gba1b mutants (Fig 3D). These results indicate that expression of GCase in the muscles of Gba1b mutants restores normal EV content and EVs may promote the spread of protein aggregates between tissues.

Tissue-specific Cer depletion is necessary for the spread of protein aggregation in Gba1b mutants

To examine whether EVs may be mediating the observed non-cell-autonomous rescue, we examined factors influencing the metabolism of EVs. We hypothesized that disruption of EV biogenesis might reduce protein aggregation in distant tissues in Gba1b mutants by decreasing production of dysregulated EVs that promote propagation of protein aggregates. To test this hypothesis, we examined knockdown of neutral sphingomyelinase (nSMase), encoding a lipid-modifying enzyme important for the formation and release of EVs [38,39]. nSMase hydrolyzes sphingomyelin, producing phosphocholine and ceramide. This enzyme has been implicated in multiple cellular functions, including inflammatory responses, reaction to lung and cardiac pathology, synaptic regulation, and release of EVs independent of ESCRT machinery [40]. We hypothesized that tissue-specific RNAi knockdown of nSMase in Gba1b mutants could reduce protein aggregation in distant tissues by reducing the production of dysregulated EVs promoting protein aggregation. However, if Gba1b mutant phenotypes are due to relative Cer depletion rather than accumulation of GlcCer, then nSMase knockdown might enhance the cell-autonomous protein aggregation in Gba1b mutants as Cer is known to trigger EV release [38]. We observed the latter, with increased accumulation of ubiquitinated proteins and Ref(2)P in thoraces (Fig 4A and 4C), suggesting that Cer depletion promotes cell-autonomous protein aggregation. However, we also observed a trend towards an increase in ubiquitinated proteins and Ref(2)P in the heads of Gba1b mutants with knockdown of nSMase expression in flight muscle (Fig 4B and 4D), suggesting that nSMase knockdown in muscle is not sufficient to reduce non-cell-autonomous spread of protein aggregates to other tissues, and may be enhancing non-cell-autonomous spread of protein aggregates. In contrast, knock down of nSMase in neuronal tissue did not modify Ref(2)P or ubiquitinated protein levels in the heads of Gba1b mutants (Fig 4E and 4F). These results suggest that nSMase-mediated release of EVs and/or influence on Cer metabolism may be important for the cell-autonomous regulation of protein aggregation in Gba1b mutant muscles but not neurons.

We next examined whether EV cargo is altered in Gba1b mutants with RNAi knockdown of nSMase in muscle. We found an even further increase in Ref(2)P in EVs isolated from Gba1b mutants with nSMase knocked down in muscle (Fig 5A). Rab11 which was elevated in Gba1b mutants was significantly increased compared to controls after knockdown of nSMase in the muscle of mutant flies (Fig 5C). There was no significant change in the levels of ubiquitinated proteins or Rab7, which were already increased in EVs from Gba1b mutants (Fig 5B and 5D). This suggests that nSMase may regulate the protein cargo and abundance of specific types of EVs.

To investigate whether the observed effects of nSMase-RNAi on cell-autonomous protein aggregation and EV cargo are due to alterations in Cer metabolism, we performed targeted lipidomic analysis to determine Cer and GlcCer levels in Gba1b mutants and controls with and without nSMase knocked down in muscle. We detected 39 Cer and 55 GlcCer species by liquid chromatography-mass spectroscopy (LC-MS) from whole bodies of 10 day old flies. Principal component analysis (PCA) of GlcCer species in Gba1b mutants and controls with and without nSMase knocked down in muscle revealed 4 distinct groups which could be distinguished by 2 main principal components (PC) (Fig 6A). Most of the variance (94.7%) was explained by PC1, separating control and Gba1b mutants. The top 15 significant GlcCer species for the loadings for PC1 and PC2 are listed in S1 Data. We found the levels of all GlcCer species in Gba1b mutants to be significantly elevated, as highlighted by the heat map displaying clustering analysis for GlcCer species in Gba1b vs control (Fig 7). All the GlcCer species were increased in Gba1b mutants by a fold-change of ≥1.5 compared to controls (Fig 6B), and 31 (56%) GlcCer species were increased over 20-fold for a False Discovery Rate (FDR) ≤ 0.05 (Fig 6B and Table 1, and S3A Fig). Consistent with PCA results, the effect of nSMase knock down in muscle on GlcCer levels had a modest effect on both control and Gba1b mutants but tended to decrease GlcCer levels (Fig 6C and 6E). Only one GlcCer species decreased by over 2-fold in Gba1b mutants expressing nSMase-RNAi in muscle compared to Gba1b mutants, and 14 GlcCer species decreased by over 1.5-fold with FDR ≤ 0.10 (Fig 6C and Table 2 and S3B Fig). Partial Least Squares–Discriminant Analysis (PLS-DA), a supervised multivariate regression analysis, was also performed to identify key GlcCer species differing between Gba1b mutants and controls, and Gba1b mutants with nSMase-RNAi in muscle vs Gba1b mutants. PCA and PLS-DA analyses resulted in very similar group separation of Gba1b mutants versus controls and Gba1b mutants expressing nSMase-RNAi in muscle versus Gba1b mutants (S2 Fig). Only 1 GlcCer species (GlcCer d16:2/24:1) in the top 15 significant GlcCer species identified by PLS-DA in Gba1b mutants expressing nSMase-RNAi in muscle versus Gba1b mutants (R² = .99, Q² = 0.98) was also in the top 15 GlcCer species differing significantly between Gba1b mutants versus controls (R² = 0.99, Q² = 0.79) (Fig 6D and 6E).

PCA of Cer species in Gba1b mutants and controls with and without nSMase knocked down in muscle revealed a distinct separation between Gba1b mutants and controls, accounting for 46.5% variance, and distinct separation between Gba1b mutants expressing nSMase-RNAi in muscle versus Gba1b mutants, accounting for 20.4% variance (Fig 8A, significant Cer species for PCA loadings in S1 Data). Heat map derived from clustering analysis of Cer species in Gba1b mutants versus controls indicates that the majority of Cer species are increased in Gba1b mutants compared to controls (Fig 9), although there is a significant subset of Cer species that are decreased. Fold-change analysis revealed that the magnitude of alterations in Cer levels between Gba1b mutants and controls was much smaller in comparison to alterations of GlcCer species in Gba1b mutants versus controls (Fig 8B). Volcano plot with adjusted p-values for FDR ≤ 0.10 revealed 36% of Cer species had greater than 1.5-fold change in Gba1b mutants versus controls, with 4 Cer species decreased and 9 Cer species increased in Gba1b mutants versus controls (Table 3 and S5A Fig). PCA and PLS-DA analyses again resulted in similar discrimination between groups when comparing Gba1b mutants versus controls, and Gba1b mutants expressing nSMase RNAi in muscle versus Gba1b mutants, (S4 Fig and S1 and S2 Data). PLS-DA of Cer species in Gba1b mutants versus controls identified two Cer species (Cer d16:2/22:1, and Cer d14:2/22:1) that were significantly reduced in Gba1b mutants versus controls (R² = 0.99, Q² = 0.80) (Fig 8D and S6 Fig). These two Cer species were also within the top 15 loadings for PC1 in PCA of Gba1b mutants versus controls (S1 Data) and reduced by greater than 1.5 fold with adjusted p-value for FDR ≤ 0.10 in Volcano plot for Gba1b mutants versus controls (Table 3 and S5A Fig). Cer d16:2/22:1 and Cer d14:2/22:1 were found to be further significantly decreased by PLS-DA of Gba1b mutants expressing nSMase-RNAi in muscle versus Gba1b mutants (R² = 0.99, Q² = 0.70) (Fig 8E and Table 4), and again found to be significant contributors to PC1 in PCA of Gba1b mutants expressing nSMase-RNAi in muscle versus Gba1b mutants (S1 Data). These results confirm that the knockdown of nSMase expression by RNAi affects Cer metabolism, as the majority of Cer species in Gba1b mutants with nSMase knocked down in muscle decreased compared to Gba1b mutants, consistent with the predicted enzymatic effect of nSMase knockdown on Cer metabolism.

Ectopically expressed GCase is trafficked within extracellular vesicles

Our finding that GBA expression in muscle normalizes Ref(2)P levels in EVs suggests that non-cell-autonomous rescue is mediated at least in part by preventing spread of aggregated proteins trafficked by EVs. However, EV-mediated trafficking of GCase could also be another contributing factor to the non-cell-autonomous rescue of protein aggregation in Gba1b mutants. To test this possibility, we investigated whether ectopically expressed GCase travels to distant tissues via EVs. Because there are no antisera that recognize Drosophila GCase, we examined ectopic expression of human GBA (hGBA) in Gba1b mutants. We previously found ubiquitous expression of WT hGBA (hGBA^WT) to be sufficient to partially rescue the shortened lifespan of Gba1b mutants [19]. Here, we found that expression of hGBA^WT in flight muscle also reduced ubiquitinated protein aggregation and Ref(2)P accumulation in Gba1b mutants in both thoraces and heads (Fig 10A–10D). EVs isolated from Gba1b mutants expressing hGBA^WT in flight muscle were found to have decreased Rab11 levels compared to Gba1b mutants (Fig 11A and 11B), confirming that human GCase (hGCase) can also revert alterations in EVs due to dGba1b deficiency. We also expressed mutant hGBA bearing the common N370S mutation (hGBA^N370S) in muscle, which failed to reduce the increased ubiquitin levels in the thoraces and heads of Gba1b mutant flies (Fig 10F and 10G). These results support a functional equivalence between Drosophila GCase and hGCase in reducing the formation of protein aggregates. Interestingly, we were also able to detect hGCase in both the thoraces and heads of Gba1b mutants expressing hGBA^WT in flight muscle (Fig 10E), indicating that hGCase itself is able to travel to distant tissues. Furthermore, we detected hGCase in EVs isolated from hemolymph of flies expressing hGBA^WT in muscle (Fig 11A). To determine whether hGCase is localized within EVs or associated on the outer surface of EVs, we added Proteinase K to the isolated EV fraction. We found hGCase to be resistant to Proteinase K digestion of the EV fraction but degraded if EVs were first treated with a detergent to disrupt vesicular membranes before Proteinase K digestion (Fig 11C). These results indicate that GCase can be incorporated into EV cargo, and that WT GCase may be trafficked by EVs to distant tissues to reduce protein aggregation.

Discussion

Many genetic influences of PD have now been identified, and much work has been focused on how these genes lead to protein aggregation through mechanisms such as protein misfolding and autophagy defects. However, none of these genes have been implicated in cell-to-cell spread of pathogenic protein aggregates, which closely correlates with clinical disease progression. Our proteomic analysis and non-cell-autonomous rescue of protein aggregation in Gba1b mutants has led us to hypothesize that GBA mutations may influence the rate of propagation of protein aggregates between neurons. Our work suggests a link between GBA mutations and faster spread of intracellular protein aggregates via a novel EV-mediated mechanism, possibly explaining the recent clinical finding that GBA mutations accelerate the progression of clinical disease. Using a Drosophila model of GBA deficiency that manifests accelerated protein aggregation, we found that expressing WT GCase in specific tissues of a GBA-deficient fly can not only rescue protein aggregation cell-autonomously and in distant tissues, but also rescue alterations in protein cargo observed in EVs isolated from Gba1b mutant hemolymph. Interestingly, ectopically expressed WT GCase itself was found within EVs of GBA-deficient flies, suggesting that the non-cell-autonomous rescue due to GCase expression is mediated by both reduction in aggregated proteins in EVs and trafficking of GCase via EVs to distant cells and tissues. Perturbing EV biogenesis through decreased expression of ESCRT-independent nSMase affected protein aggregation in local tissues in a tissue-dependent manner, and further decreased a subset of Cer species already reduced in Gba1b mutants. Interestingly, this subset of Cer species is known to be enriched in EV membranes [41]. Together, these findings suggest that mutations in GBA result in the accelerated spread of protein aggregates through changes in cellular lipid composition and dysregulation of proteins trafficked by EVs.

Although our model of GBA mutations promoting spread of protein aggregates via EVs is novel, the idea that proteostasis can be maintained in a non-cell-autonomous fashion is well supported in the literature. For example, in C. elegans, misfolded α- synuclein accumulating in endo-lysosomal vesicles was found to be transmitted from muscle to the hypodermis, a nearby tissue, for degradation [42]. It is possible that a non-cell-autonomous mechanism is necessary because certain tissues may be more efficient in reducing protein aggregation. This has been previously described, where overexpression of FOXO in Drosophila muscle decreased aging-related protein aggregates in muscle as well as brain and other distant tissues, but FOXO overexpression in adipose tissue was unable to prevent protein aggregation in muscle [43]. In our model, overexpressing dGba1b in Drosophila muscle or neuronal tissue prevented accumulation of protein aggregates throughout the organism, however overexpression of WT GCase in midgut and fat body did not significantly reduce protein aggregation in the brain (S8 Fig). These discrepancies could be due to tissue-specific biogenesis of EVs, which could depend on factors such as metabolic rate or endovesicular trafficking. Although dGba1b is expressed in all tissues, a second homologue of human GBA1, dGba1a, is expressed only in the midgut. Our Gba1b mutants retain ~25% expression of dGba1a [19]. Deficiency of dGba1a was found to extend lifespan [20] and does not result in significant accumulation of GlcCer [21], suggesting that there can be significant tissue-specific differences in function for GCase that could influence EV biogenesis.

Our unexpected results from perturbation of EV biogenesis suggest that the EV-mediated regulation of protein aggregation is tissue-specific and complex. Because an increase in EV-intrinsic proteins and alteration of protein cargo were observed in Gba1b mutants [23], we anticipated that genetic perturbations decreasing the biogenesis of EVs might rescue protein aggregation non-cell-autonomously by reducing the production of dysregulated EVs. However, decreased expression of ESCRT-independent nSMase in muscle did not rescue protein aggregation in heads, suggesting that a tissue-specific decrease in biogenesis of dysregulated EVs is not sufficient to reduce protein aggregation in the rest of the organism, and the cargo of EVs may need to be corrected to reduce spread of protein aggregation. In contrast, decreased expression of nSMase in the nervous system had no effect on protein aggregation in the head. This difference in outcome in perturbation of EV biogenesis in muscle and neurons could be due to cell-specific compensatory mechanisms or intrinsic metabolic demands and solicits further investigation.

A possible explanation for why decreased muscle expression of nSMase enhanced cell-autonomous protein aggregation and EV protein cargo alterations is that both GCase and nSMase enzymatically produce Cer. If GCase-deficient phenotypes are dependent on a relative reduction in Cer, decreased nSMase expression could exacerbate Gba1b mutant phenotypes. Indeed, lipidomic analysis of alterations in Cer metabolism due to nSMase knockdown revealed a further decrease in a subset of Cer species that were already significantly decreased in Gba1b mutants compared to controls. The further reduction in Cer species due to nSMase knockdown correlates with enhancement of cell-autonomous protein aggregation and EV cargo alterations, suggesting that accelerated protein aggregation in Gba1b mutants is mediated by Cer deficiency rather than GlcCer accumulation, as nSMase knockdown had a much more modest effect on the significantly increased levels of GlcCer species in Gba1b mutants compared to controls.

Cer has been implicated in the composition and biogenesis of EVs, and nSMase knockdown further altered EV cargo in Gba1b mutants, suggesting that decreased Cer levels may directly influence EV biogenesis in Gba1b mutants. However, Cer species were not globally decreased, suggesting that the regulation of Cer metabolism is complex and may be more dependent on specific Cer species. Interestingly, only 1 of the 9 Cer species significantly increased in Gba1b mutants versus controls had a monounsaturated fatty acyl group, while all 5 of the Cer species significantly decreased in Gba1b mutants versus controls had a monounsaturated fatty acyl group, suggesting GBA influences the metabolism of specific subset of Cer species that may be implicated in Gba1b mutant phenotypes. This subset of Cer species is enriched in species with long chain monounsaturated fatty acyl chains. Interestingly, lipids with monounsaturated fatty acyl groups are an abundant component in mammalian exosome membranes [40]. Investigating the alterations in lipid composition of EVs resulting from GCase deficiency and nSMase knockdown will be important in elucidating the role of Cer metabolism in Gba1b mutant phenotypes.

Our work suggests that GCase deficiency influences EV biogenesis to promote faster propagation of pathogenic protein aggregates throughout the tissues of an organism, which may be a compensatory response to cell-autonomous lysosomal stress. In the initial characterization of our Drosophila GBA-deficient model we found accelerated insoluble ubiquitinated protein aggregates, accumulation of Ref(2)P, and oligomerization of ectopically expressed human α-synuclein in Gba1b mutants, suggesting an impairment in lysosomal degradation [19]. A similar GBA-deficient Drosophila model also found evidence of lysosomal dysfunction, including enlarged lysosomes in GBA-deficient brains [20]. However, our proteomic analysis of Gba1b mutants did not support a profound impairment in autophagy, but instead suggested dysregulation of EVs with altered protein cargo which could be suppressed locally with knockdown of genes encoding ESCRT machinery important for EV biogenesis [23]. Based on these results, we believe that our initial observations of increased insoluble ubiquitinated proteins and Ref(2)P in Gba1b mutants are due to lysosomal stress. One possible explanation for our proteomic findings is that there may be a compensatory increase in EV biogenesis and packaging of autophagy substrates within EVs for discard outside of the cell in Gba1b mutants. Such an increase may have prevented us from detecting defects in autophagy. Upregulation of EV biogenesis may be cell-autonomously neuroprotective in the setting of lysosomal stress, particularly in aggregation-prone neurodegenerative diseases such as PD [44]. It was recently demonstrated in a human neuronal cell culture model of PD that inhibiting macroautophagy protects against α-synuclein-induced cell death by promoting the release of α-synuclein-containing EVs [45]. However, it remains possible that upregulating EV biogenesis may relieve lysosomal stress within cells containing aggregate-prone proteins, while simultaneously promoting the spread of protein aggregates between cells and throughout the organism.

Our work suggests a novel mechanism for GBA in reducing the spread of pathogenic protein aggregation from cell-to-cell via regulation of EV protein cargo, but many key questions remain. To better understand the progression of neurodegenerative diseases, we must uncover the mechanisms by which GCase deficiency alters EV protein content and biogenesis, identify the specific changes in EVs facilitating propagation of pathogenic protein aggregates, and determine how these changes influence recipient cells internalizing dysregulated EVs. GCase is a critical enzyme in ceramide metabolism, hydrolyzing glucosylceramide into glucose and ceramide. Ceramides are a key component of EV membranes, and alterations in ceramide metabolism due to GCase deficiency may directly influence EV biogenesis and protein cargo trafficked via EVs. Further studies using this Drosophila model and mammalian cell culture models should better elucidate how GCase deficiency alters the protein cargo of EVs to induce propagation of pathogenic protein aggregates, as well as whether endogenous GCase is enzymatically functional when trafficked to distant tissues via EVs. Understanding this mechanism could have broad implications in understanding the pathogenesis of aggregate-prone neurodegenerative diseases and reveal new therapeutic targets to slow or halt disease progression.

Materials and methods

Drosophila strains and culture

Fly stocks were maintained on standard cornmeal-molasses food at 25°C. The Gba1b homozygous null mutant (Gba1b^ΔTT), isogenic control (Gba1b⁺), and UAS-dGba1b alleles have been previously described [19]. All other strains and alleles were obtained from the Bloomington Stock Center: elav-GAL4 (458); Act88F-GAL4 (38459); DMef-GAL4 (27390); UAS-nSMase-RNAi (36759); Gba1b^MB03039(23602). In Figs 4–7, we used the following genotypes for the experiments involving the nSMase-RNAi transgene: control = Gba1b⁺/Gba1b^MB03039; Gba1b mutant = Gba1b^ΔTT/Gba1b^MB03039. This combination of Gba1b mutant alleles, which we used for ease of recombination with the transgene, produce the same biochemical abnormalities found in Gba1b^ΔTT homozygotes [23].

Lifespan analysis

Longevity assays were conducted at 25°C. Groups of 10–20 age-matched flies were collected at 0–24 hours old and transferred to fresh standard food every 2–3 days. The number of dead flies was recorded during each transfer. Transfers were continued until all flies died. Kaplan-Meier lifespan curves were generated using Stata (StataCorp, College Station, TX), and analyzed by Cox proportional hazard models for statistically significant differences in survival between tested genotypes.

Preparation of Triton-soluble and insoluble fractions

Tissues from 10-day-old flies (6 females and 6 males per sample) were homogenized in Triton lysis buffer (50 mM Tris-HCl pH 7.4, 1% Triton X-100, 150 mM NaCl, 1 mM EDTA), and then spun at 15,000 x g for 20 min. The detergent-soluble supernatant was collected and mixed with an equal volume of 2x Laemmli buffer (4% SDS, 20% glycerol, 120 mM Tris-Cl pH 6.8, 0.02% bromophenol blue, 2% β-mercaptoethanol), and the same buffer was used to resuspend the Triton-insoluble pellet. All samples were boiled for 10 minutes. The Triton-insoluble protein extracts were then cleared of debris by centrifugation at 15,000 x g for 10 minutes, followed by collection of the supernatant. At least three independent experiments were performed.

Western blotting

Proteins were separated by SDS-PAGE on 4%-20% MOPS-acrylamide gels (GenScript Express Plus, M42012) and electrophoretically transferred onto Immobilon PVDF membranes (Fisher, IPVH00010). Immunodetection was performed using the iBind Flex Western Device (Thermo Fisher, SLF2000). Antibodies were used at the following concentrations: 1:10,000 mouse anti-Actin (Chemicon/Bioscience Research Reagents, MAB1501), 1:250 mouse anti-Rab11 (BD Transduction Laboratories, 610657), 1:50 mouse anti-Rab7 (DSHB Rab7), 1:200 rabbit anti-Ref(2)P (Abcam, ab178440), 1:500 mouse anti-ubiquitin (Santa Cruz, sc-8017), 1:10,000 mouse anti-tubulin (DSHB 12G10), 1:1500 rabbit anti-human GBA1 (Sigma, G4171), 1:800 mouse anti-Cnx99A (DSHB, Cnx99A 6-2-1). HRP secondary antibodies were used as follows: 1:500 to 1:1000 anti-mouse (BioRad, 170–6516) and 1:500 to 1:1000 anti-rabbit (BioRad, 172–1019). Signal was detected using Pierce ECL Western Blotting Substrate (Fisher, 32106). Densitometry measurements of the western blot images were performed using Fiji software [46]. (See uncropped western blot images in S4 Data.) For homogenates, signal was normalized either to Actin or Ponceau S [47,48]. Normalized western blot data were log-transformed when necessary to stabilize variance before means were compared using one-way ANOVA and Student t-test. Each experiment was performed at least three times.

Immunohistochemistry

10-day-old adult brains were dissected in cold Schneider’s Drosophila medium (Thermo Fisher, 21720), and fixed in 4% paraformaldehyde/PBS for 30 min. Samples were washed in 0.1% Triton X-100/PBS. Fixed brains were stained with 1:200 mouse anti-poly Ubiquitin FK2 (Enzo, BML-PW8810), then anti-mouse Alexa 488 (1:200) and were mounted using ProLong Gold anti-fade medium (Molecular Probes, P10144).

Extraction of hemolymph and preparation of EV fractions

Hemolymph was obtained from 30 flies (15 males and 15 females, 10 to 11 days old) per sample. All flies were frozen with liquid nitrogen and decapitated by vortexing. Frozen flies were placed in a 1.7-mL tube containing a volume of PBS scaled to the number of flies used (2 μL/fly) and thawed for 5 min at room temperature. The tubes were then centrifuged at 5000 x g for 5 min at 4°C, after which the extracted hemolymph (supernatant) was centrifuged for 30 min at 10,000 x g at 4°C to remove cell debris and the cell-free supernatant was collected. The supernatant was then filtered with Ultrafree 0.22 μm spin filters (Fisher, UFC30GV0S) and centrifuged at 3000 x g for 5 min at 4°C; this was the EV fraction. In order to obtain whole-fly homogenate from the same animals used for collection of hemolymph, the bodies from four flies (2 males and 2 females) were homogenized in RIPA buffer (150mM NaCl, 1% Nonidet P-40, 0.5% Sodium deoxycholate, 0.1% SDS, 50mM Tris pH 8, diluted 1:1 with ddH₂0), centrifuged at 10,000 x g at 4°C for 5 min, and then the supernatant was transferred to a new tube. An equal volume of 2x Laemmli buffer (4% SDS, 20% glycerol, 120 mM Tris-Cl pH 6.8, 0.02% bromophenol blue, 2% β-mercaptoethanol) was added to the EV fractions and also to the whole-fly protein homogenates, and all samples were boiled for 10 min and then stored at −80°C. The experiment was repeated at least three times.

To determine if proteins were contained within EVs, before adding the Laemmli buffer, the EV fractions were incubated with PBS for 5 min at room temperature and then 0.5 μg/μL Proteinase K was added for 30 min at 4°C to digest external proteins or with 1% sodium dodecyl sulfate (SDS) for 5 min at room temperature and then 0.5 μg/μL Proteinase K was added for 30 min at 4°C to disrupt vesicular membranes and digest proteins. After incubation, Laemmli buffer was added and samples were boiled preparing them for western blotting.

Lipidomic analysis

Fly tissue samples were prepared for LC-MS analysis by adding an equal volume of zirconium oxide beads to each sample of fly tissue (10 male and 10 female whole 10-day old flies). 200 μL water was added to each sample and homogenized using a Next Advance Bullet Blender at 4°C. 100 μL of homogenate was set aside for each sample for protein quantification. 50 μL of the isotope mix (Lipidyzer internal standard 1 μM d9-CER and 0.5 μM d9-HCER), 575 μL methyl tert-butyl ether (MTBE), and 160 μL methanol were added to the remaining 100 ul of each sample. All samples were vortexed and shaken for 30 minutes at room temperature at a 500 RPM setting. An additional 200 μL of water were added. The samples were centrifuged for 3 minutes at 12,000 x g at room temperature. The upper layers were transferred into glass culture tubes using glass pipettes. A second extraction was performed by adding 300 μL MTBE, 100 μL methanol, and 100 μL water. The samples were shaken for 5 minutes and centrifuged at the same settings as before. The upper layers were combined into the respective glass culture tubes and dried under a gentle stream of nitrogen. The samples were reconstituted in 250 μL of mobile phase B (60% acetonitrile, 40% isopropyl alcohol, 0.2% formic acid), and transferred to LC vials with inserts for LC-MS analysis.

Lipids were analyzed using a Shimadzu Nexera X2 HPLC (Shimadzu, Japan) coupled to an AB Sciex QTRAP 5500 hybrid triple quadrupole/linear ion trap mass spectrometer (AB Sciex LLC, USA) operating in positive electrospray ionization (ESI) mode. 39 ceramides and 55 glucosylceramides were monitored by multiple-reaction-mode (MRM) in a single batch run.

Lipidomic data were exported as csv files, grouped according to Drosophila strain and culture, and placed into MetaboAnalyst 4.0 web server (https://www.metaboanalyst.ca) for data processing and statistical analysis [49]. (See all lipidomic data in S3 Data.) Data was normalized by mass per sample, log-transformed and auto-scaled. The data were evaluated by univariate analysis methods (Fold Change analysis, T-tests, and Volcano plot) and multivariate analysis methods (principal component analysis and partial least square–discriminant analysis), and hierarchical clustering analysis. The False Discovery Rate (FDR), based on the Benjamini-Hochberg procedure, was set to <0.1 and used to select important Cer and GlcCer species. The FDR-adjusted p-value of <0.10 was used to control the number of false positives for multiple comparisons.

Supporting information

S1 Fig [a]
Representative images for .

S2 Fig [a]
Scores plots for PCA and PLS-DA for GlcCer species.

S3 Fig [a]

S4 Fig [a]
Scores plots for PCA and PLS-DA for Cer species.

S5 Fig [a]

S6 Fig [tif]
Box plots of normalized values of Cer species that were significantly altered in both vs controls, and ; RNAi vs , as determined by VIP score from PLS-DA, and within top 15 species of PC1 loadings for PCA of vs control and ; RNAi vs .

S7 Fig [ub]
Representative images for .

S8 Fig [a]
Midgut and fat body expression of does not rescue protein aggregation in mutants.

S1 Table [pdf]
Summary of Lifespans in .

S2 Table [pdf]
Summary of Lifespans in .

S1 Data [xlsx]
PCA loadings.

S2 Data [xlsx]
PLS-DA loadings.

S3 Data [xlsx]
Targeted lipidomic results.

S4 Data [pdf]
All uncropped western blots.